1136/bjophthalmol-2016-308948
Clinical science
the Queen Elizabeth II Health Sciences Centre in Halifax, Nova enough postoperative VF information for these patients.
Scotia, Canada. To ensure that the MD rates were based on a Accordingly, we identified 180 surgical eyes from 180 patients
reasonable number of VF examinations, only patients with five who met all the inclusion and exclusion criteria, which were
or more SITA-Standard 24–2 Humphrey Field Analyzer (Carl matched by baseline MD to 180 medically treated eyes.
Zeiss Meditec, Dublin, California, USA) VF examinations after In the surgically treated group, 86 eyes (47.8%) underwent
surgery were included in the analysis. If both eyes of a patient trabeculectomy alone and 94 eyes (52.2%) underwent combined
fulfilled the inclusion criteria, one eye was randomly selected cataract extraction and trabeculectomy. The mean (±SD) time
for analysis. The first postsurgery VF test was defined as the from surgery to first (baseline) VF was 0.8 (±1.0) years.
baseline examination for these patients. The mean baseline MD (±SD) in the surgically and medically
Laser suture lysis, bleb needling, antimetabolite subconjuncti- treated groups was −8.72 (±5.24) dB and −8.71 (±5.22) dB,
val injection, topical and/or systemic glaucoma medication and respectively ( p=0.38). The mean (SD) unsigned difference in
surgical revision were used during the postoperative period baseline MD within matched pairs of surgically and medically
when clinically indicated and were not considered exclusion cri- treated patients was 0.04 (±0.09) dB. Demographics of the
teria. However, eyes that underwent a subsequent glaucoma sur- patients and characteristics of the study eyes are shown in table 1.
gical procedure (second trabeculectomy, glaucoma drainage Surgically treated patients had significantly more VF tests com-
device implantation, cyclophotocoagulation or any other surgi- pared with medically treated patients (10.8 and 8.4, respectively,
cal glaucoma procedure) were excluded from the study. p<0.01). There was also a trend for longer follow-up time in the
surgically treated eyes compared with the medically treated eyes
Medically treated patients (7.4 years and 6.8 years, respectively, p=0.05).
Each surgically treated patient was matched by baseline MD to There was no statistically significant difference in the mean
a patient treated by topical medical therapy (with or without MD slopes between the two groups: −0.22 (±0.55) dB/year
the additional use of laser trabeculoplasty) who had at least five postoperatively in the surgically treated patients, and −0.08
SITA-Standard 24–2 VF tests. These medically treated patients (±1.10) dB/year in the medically treated patients ( p=0.13, 95%
were extracted from a large database of patients with glaucoma CI of the mean difference: −0.31 to 0.04). Figure 1 shows the
treated in our centre1 and were managed clinically throughout distribution of the MD slopes in the two groups. In the medical
the whole follow-up, without ever having a surgical glaucoma and surgical group, respectively, 7 (3.9%) and 17 (9.4%)
procedure. Similarly, in this group, if both eyes from the same patients were classified as fast progressors ( p=0.05). There was
patient fulfilled the inclusion and exclusion criteria, only one no statistically significant difference ( p=0.10) in the MD slopes
eye was randomly selected for analysis. of the patients who underwent combined cataract extraction
Patients in either group with any other ocular or non-ocular and trabeculectomy (−0.27±0.50 dB/year) or trabeculectomy
disease known to cause VF defects were excluded. No patient alone (−0.17±0.60 dB/year). Figure 2 shows the MD slopes as a
contributed with both eyes in the study (with one in each function of the baseline MDs in both groups, suggesting no dif-
group). ference between the groups in any range of severities.
Surgically treated patients were using a mean of 2.5 (±1.0)
Data collection topical medications before surgery, which was reduced to 0.9
Parameters collected from the clinical charts included age, (±1.1), at the last follow-up visit post surgery ( p<0.01, paired
gender, type of glaucoma, refraction, axial length, preoperative t-test). Their mean IOP before surgery was 19.8 (±7.5) mm Hg,
glaucoma laser procedures, lens status, visual acuity, IOP and which was reduced to 12.0 (±4.4) mm Hg at the last follow-up
ocular hypotensive medications. In the surgically treated eyes, visit post surgery ( p<0.01, paired t-test). The mean and
preoperative IOP was calculated as the mean of the three most minimum IOPs were significantly lower and IOP fluctuation
recent values measured prior to surgery; in this group we (maximum IOP−minimum IOP) was significantly higher in the
excluded from the analysis the IOP values measured during the surgically treated group postoperatively compared with the med-
first 2 months after surgery, due to the possibility of large IOP ically treated group. Details of IOP parameters in the two
fluctuations in the immediate postoperative period. groups can be found in table 2.
The mean IOP of patients with fast progression was 12.7
Statistical analysis (±2.4) mm Hg and 17.6 (±3.9) mm Hg in the surgically
The rate of MD change was calculated for each patient with
robust regression analysis. Robust regression is a technique that
provides slope estimates that are more resistant to outlier obser-
vations.11 Patients with MD slopes < −1 dB/year were classified Table 1 Demographics and clinical information from the surgically
as fast progressors.1 We compared the average MD slopes in and medically treated patients (p values determined by paired
both groups with a paired t-test and the number of patients t-test)
with fast progression using the Fisher’s exact test.
Surgically treated Medically treated
Statistical analysis was carried out in the open-source pro- patients patients
gramming language R. (R Foundation for Statistical Computing, n=180 patients n=180 patients
V.3.2.0), using package ‘robustbase’ (V.0.92–3). (eyes) (eyes)
Mean (±SD) Mean (±SD)
RESULTS Baseline MD (dB) −8.72 (±5.24) −8.71 (±5.22) p=0.38
During the 14-year-period we evaluated, trabeculectomy and/or
Age (years) 69.7 (±10.7) 67.2 (±14.4) p=0.08
combined cataract extraction and trabeculectomy surgery were
Number of visual 10.8 (±4.3) 8.4 (±3.7) p<0.01
performed in 4135 eyes. Due to the tertiary nature of our field tests
clinic, serving a very large geographical referring area, the vast Follow-up (years) 7.4 (±2.9) 6.8 (±3.1) p=0.05
majority of our surgical patients are followed after surgery by
dB, decibels; MD, mean deviation.
their referring ophthalmologist and, therefore, we did not have
2 Baril C, et al. Br J Ophthalmol 2016;0:1–5. doi:10.1136/bjophthalmol-2016-308948
Clinical science
(n=17) and medically (n=7) treated patients, respectively. In patients: −0.22 dB/year and −0.08 dB/year, respectively.
the surgical cohort, mean IOP in the non-fast progressors However, more patients in the surgical group showed fast VF
(n=163) was 12.1 (±3.5) mm Hg, similar to the IOP observed progression (rates worse than −1 dB/year) than in the medically
in the fast progressors ( p=0.46). There was no statistically sig- treated group (17 and 7, respectively), a difference that
nificant difference in baseline MDs between the fast progressors approached statistical significance.
compared with the non-fast progressors in both the surgically Although it is difficult to compare across different studies,
and medically treated patients ( p=0.82 and 0.29, respectively). our results in the surgical group are in the range of MD loss
Results are summarised in table 3. reported after glaucoma surgery: Bertrand et al5 and Folgar
et al6 observed a mean MD change after surgery of −0.16 dB/
DISCUSSION year in 52 eyes and −0.49 dB/year in 74 eyes, respectively. In
The main goal of our study was to evaluate if successful trabecu- eyes with progressive normal tension glaucoma, Iverson et al7
lectomy or combined cataract extraction and trabeculectomy and Shigeeda et al8 reported mean postoperative MD change of
would prevent further VF progression. In order to avoid the −0.25 dB/year and −0.44 dB/year respectively, compared with
possibility of a regression to the mean effect, we compared our −1.05 dB/year preoperatively in both cohorts. In a recent study
rates of progression post surgery with a control group matched by Caprioli et al,10 the mean decay rate (measured by the VF
by baseline MD and composed of patients who were treated index) in patients who underwent trabeculectomy slowed from
medically, without ever having glaucoma surgery. A similar −2.4±9.3%/year preoperatively to −0.6±13.1% after surgery.
approach was used by Caprioli et al10 in a recently published Elevated IOP has been confirmed as a significant risk factor
manuscript. for glaucoma progression in many landmark studies.12–17 In our
We found that the rate of VF progression in the surgically study, as expected, the surgically treated eyes had significantly
treated patients was similar to that observed in medically treated lower mean IOP postoperatively than the medically treated
Table 3 Baseline mean deviation (MD) and mean follow-up intraocular pressure (IOP) characteristics in fast progressors (MD < −1 dB/year)
compared with non-fast progressors in surgically and medically treated patients (p values determined with Mann-Whitney U test)
Baseline MD (dB) IOP (mm Hg)
Mean (±SD) Mean (±SD)
Surgically treated group Non-fast progressors −8.76 (±5.25) p=0.82 12.1 (±3.5) p=0.46
n=163 eyes
Fast progressors −8.34 (±5.23) p=0.29 12.7 (±2.4)
n=17 eyes
Medically treated group Non-fast progressors −8.64 (±5.24) 16.6 (±3.5) p=0.73
n=173 eyes
Fast progressors −10.38 (±4.47) 17.6 (±3.9)
n=7 eyes
dB, decibels.